INDEX

Explanations

phrases emphasizing negation or absence

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

IST

-0.45

 Unterneh

-0.43

ly

-0.42

charAt

-0.41

 estudi

-0.40

 Bergmann

-0.39

 recherchez

-0.39

 acuer

-0.39

 pasada

-0.38

ukary

-0.38

POSITIVE LOGITS

no

1.14

No

1.08

No

1.04

NO

0.91

aucun

0.84

 none

0.81

 nessun

0.79

aucune

0.79

no

0.74

NO

0.74

Activations Density 0.154%

phrases emphasizing negation or absence

No Comments

No Known Activations